cleanup the plots

analytics.py

@@ -612,9 +612,14 @@ def _plot_company_bars(
     text_in = [value_fmt.format(v) for v in values]
     text_colors = [f"rgba(255,255,255,{a:.3f})" for a in text_alphas]
 
+    # Use integer x-positions so duplicate display labels (e.g. two models that
+    # share the same truncated short name) don't collapse into a single stacked
+    # bar. Tick labels are mapped back onto these positions below.
+    x_positions = list(range(len(labels)))
+
     fig = go.Figure(
         go.Bar(
-            x=labels,
+            x=x_positions,
             y=values,
             customdata=list(model_ids),
             marker=dict(
@@ -647,6 +652,9 @@ def _plot_company_bars(
             font=dict(size=18),
         ),
         xaxis=dict(
+            tickmode="array",
+            tickvals=x_positions,
+            ticktext=list(labels),
             tickangle=-40,
             automargin=True,
             showgrid=False,
@@ -666,12 +674,12 @@ def _plot_company_bars(
     return fig
 
 
-def plot_intelligence_bars(
+def plot_accuracy_bars(
     df: pd.DataFrame,
     top_n: int = 10,
 ) -> go.Figure:
     """
-    Ranked vertical bars of the far-field "Intelligence" score (`ff_score`).
+    Ranked vertical bars of the far-field accuracy score (`ff_score`).
 
     Mirrors the Artificial Analysis style: top models sorted descending, value
     inside the bar, rounded corners, one color per HF org.
@@ -692,7 +700,7 @@ def plot_intelligence_bars(
         labels=labels,
         values=[float(v) for v in d["ff_score"].tolist()],
         model_ids=d["model_id"].tolist(),
-        title="Intelligence",
+        title="Accuracy",
         subtitle="Far-field accuracy score · Higher is better",
         value_fmt="{:.0f}",
         hover_value_fmt="%{y:.2f} / 100",
@@ -707,7 +715,7 @@ def plot_speed_bars(
     """
     Ranked vertical bars of throughput (`eval_rtf`, × realtime).
 
-    Companion to `plot_intelligence_bars`: same color-per-company convention,
+    Companion to `plot_accuracy_bars`: same color-per-company convention,
     same in-bar value labels, rounded corners.
     """
     if df.empty or "eval_rtf" not in df.columns:

app.py

@@ -288,7 +288,7 @@ def _analytics_initial():
     line_default = [k for k in live_keys if k in avail] or list(avail)
     sel = ids[: min(5, len(ids))]
     tn = 10
-    fig_intel = analytics.plot_intelligence_bars(df, top_n=tn)
+    fig_intel = analytics.plot_accuracy_bars(df, top_n=tn)
     fig_speed = analytics.plot_speed_bars(df, top_n=tn)
     fig_hm = analytics.plot_scenario_heatmap(df, line_default, top_n=tn)
     fig_pareto = analytics.plot_pareto_frontier(df)
@@ -318,7 +318,7 @@ def _analytics_apply(models, line_keys, top_n):
     if not lk:
         lk = avail
     tn = int(top_n) if top_n else 10
-    fig_intel = analytics.plot_intelligence_bars(df, top_n=tn)
+    fig_intel = analytics.plot_accuracy_bars(df, top_n=tn)
     fig_speed = analytics.plot_speed_bars(df, top_n=tn)
     fig_hm = analytics.plot_scenario_heatmap(df, lk, top_n=tn)
     fig_pareto = analytics.plot_pareto_frontier(df)
@@ -811,7 +811,7 @@ with gr.Blocks(title=APP_TITLE, theme=_theme, css=LEADERBOARD_CSS) as demo:
             gr.Markdown(
                 "Charts support standard Plotly interaction (legend, zoom, pan, hover).\n\n"
                 "**Leaderboard order** is by **Average WER** (mean across populated live scenarios; lower is better).\n\n"
-                "**Intelligence** ranks the top N models by far-field score (WER-only FFAS, higher is better). "
+                "**Accuracy** ranks the top N models by far-field score (WER-only FFAS, higher is better). "
                 "**Speed** ranks the top N models by RTFx (audio sec / inference sec, higher is better). "
                 "Bars are colored per company (e.g. all NVIDIA models share one color).\n\n"
                 "**WER heatmap** shows WER (%) by model and scenario for the top N models (sorted by Average WER); "
@@ -846,7 +846,7 @@ with gr.Blocks(title=APP_TITLE, theme=_theme, css=LEADERBOARD_CSS) as demo:
             )
             an_apply = gr.Button("Apply / refresh charts", variant="primary")
             with gr.Row():
-                an_plot_intelligence = gr.Plot(label="Intelligence")
+                an_plot_intelligence = gr.Plot(label="Accuracy")
                 an_plot_speed = gr.Plot(label="Speed")
             an_plot_compare = gr.Plot(label="WER heatmap")
             gr.Markdown(